Activity-based probes and chemical proteomics uncover the biological impact of targeting HMG-CoA Synthase 1 in the mevalonate pathway

Abstract

Mevalonate is a precursor for essential metabolites, such as isoprenoids and sterols. Its synthesis starts with 3-hydroxy-3-methylglutaryl-coenzyme A synthase 1 (HMGCS1) producing HMG-CoA, which is then converted to mevalonate by HMG-CoA reductase (HMGCR), a target of statins. Cancer cells often upregulate enzymes in the mevalonate pathway to meet their metabolic demands, leading to the development of inhibitors targeting several enzymes in this pathway. However, current inhibitors have not yet shown significant anticancer activity. While HMGCS1 has unique biochemical properties that distinguish it from other mevalonate pathway enzymes, the effects of inhibiting HMGCS1 have not been thoroughly investigated. Here, we present a set of chemical probes that enable us to systematically assess the proteome-wide selectivity and potency of Hymeglusin, the primary inhibitor of HMGCS1 used in the field, confirming it as a useful tool for short-term HMGCS1 inhibition. Inhibiting HMGCS1 with Hymeglusin causes proteome changes that are nearly identical to those caused by inhibiting HMGCR or degrading HMGCS1. Accordingly, simultaneously targeting HMGCS1 and HMGCR effectively suppresses the growth of statin-resistant cells and xenograft models, without increasing the risk of side effects. Finally, we find that while Hymeglusin is a valuable tool for short-term mechanistic studies, its usefulness is limited for long-term efficacy studies due to its poor stability in serum. Together, this study highlights the biological implications of targeting HMGCS1 as monotherapy or in combination with statins, and caution is required when using Hymeglusin as a tool.

Keywords: HMGCR; HMGCS1; activity-based probe; chemical proteomics; hymeglusin; mevalonate; statin.

Figure 1

Hymeglusin-fluorescence conjugates serve as activity-based probes for HMGCS1.A, a schematic of the HMGCS1 and Hymeglusin (top) and the zoomed-in cocrystal structure of the Hymeglusin-bound HMGCS1 from Brassica juncea (bottom, PDB: 2F9A). B, chemical structures of the Hymeglusin-fluorescein (HG-FL) and Hymeglusin-tetramethylrhodamine probes (HG-TMR). C, comparative in-gel fluorescence assay of WT or a catalytically inactive HMGCS1 mutant (C129A, CA) confirms the labeling of the catalytic cysteine of HMGCS1 with HG-FL. D, top: scheme of the HG-FL competition reaction in vitro. Bottom: incubating recombinant HMGCS1 (1 μg) with increasing concentrations of Hymeglusin for 30 min results in a decrease of HG-FL–labeled HMGCS1. E, the labeling of HMGCS1 by HG-FL was reversed after boiling at 95 °C for at least 20 min in a buffer containing 2% LDS and 50 mM DTT. SDS-PAGE resolved HMGCS1 before the in-gel fluorescence analysis. The stability of the HG-FL probe itself in the given conditions was assessed by dot blot in parallel and showed no change. F, HEK293T WT and HCT116 knock-in (KI) (HMGCS1-FKBP12^F36V or HMGCS1-mEGFP) cells were treated with Hymeglusin (4 μM for WT, 0.5 μM for KI, 2 h) or left untreated, followed by cell lysis and incubation with HG-FL or HG-TMR for 1 h. After in-gel fluorescence analysis, the gels were transferred to the polyvinylidene fluoride membrane for immunoblotting using an anti-HMGCS1 antibody. G, HMGCR inhibition by statins leads to the accumulation of HMG-CoA in cells. This results in the nonenzymatic modification of several residues in the FASN active site by the HMG moiety, which can be detected by the anti-HMG antibody (top). HCT116 cells expressing endogenous HMGCS1-FKBP12^F36V were treated with the indicated chemicals for 24 h, followed by immunoblotting analysis using anti-HMG, anti-HMGCR, anti-HMGCS1, and anti-tubulin antibodies (bottom). FASN, fatty acid synthase; HMGCS1, 3-hydroxy-3-methylglutaryl-coenzyme A synthase 1; LDS, lithium dodecyl sulfate.

Figure 2

Hymeglusin is a highly selective inhibitor of HMGCS1 in human cells.A, the structure of the Hymeglusin-biotin probe is shown at the top, and the workflow of competitive affinity purification mass spectrometry (AP-MS) using Hymeglusin-biotin combined with TMT-6 plex methods is shown at the bottom. B, HEK293T cells were either treated with HG (1 μM, 30 min) or dimethyl sulfoxide, followed by further incubation with the in-house HG-biotin probe (1 μM, 2 h). After streptavidin beads enrichment, the eluates were analyzed through tandem-mass-tag–based proteomic analysis. The proteomic data are presented as a volcano plot of the –log₁₀-transformed p value versus the log₂-transformed ratio of dimethyl sulfoxide/Hymeglusin pre-treated cells. n = 3 biological replicates. p-values were calculated by two-sided Welch’s t test (adjusted to 1% FDR for multiple comparisons, S0 = 0.585). C, HEK293T cells were treated as described in panel A. After streptavidin enrichment, the eluates were subjected to immunoblotting using the anti-HMGCS1 antibody. Rep: replicate. D, HEK293T cells were treated with increasing concentrations of Hymeglusin for 2 h, followed by HG-FL treatment and in-gel fluorescence analysis. The same extracts were then transferred to a polyvinylidene fluoride membrane to probe for HMGCS1 using immunoblotting. The average of two replicates is shown on the right. FDR, false discovery rate; HG, Hymeglusin; HG-FL, Hymeglusin-fluorescein; HMGCS1, 3-hydroxy-3-methylglutaryl-coenzyme A synthase 1; TMT, tandem mass tag.

Figure 3

Perturbing HMGCS1 by hymeglusin and HMGCR by simvastatin leads to identical global proteome changes.A, left panel: a schematic illustrating the mevalonate pathway flux. Three perturbation approaches—inhibition or degradation of HMGCS1 and inhibition of HMGCR—are compared for their effects on global proteome changes. Right panel: HCT116 HMGCS1-FKBP12^F36V knock-in cells were treated with dTAGv1 (100 nM, for HMGCS1 degradation), Hymeglusin (5 μM, for HMGCS1 inhibition), and simvastatin (10 μM, for HMGCR inhibition) for 48 h. The cell extracts were then probed with the indicated antibodies to check the cellular response. B, the 16-plex TMTpro approach was utilized to compare the global proteome changes after 24 h of the specified small molecule treatment. C, HEK293T cells treated as described in panel B were probed with the indicated antibodies as a quality control step prior to proteomic analysis. D, analysis of the TMTpro-plex data is presented as a volcano plot of the −log₁₀-transformed p value versus the log₂-transformed ratio of Hymeglusin-treated/untreated (UT) conditions for HCT116 HMGCS1-FKBP12^F36V knock-in cells. p values were calculated by two-sided Welch’s t test (adjusted to 1% FDR for multiple comparisons, S0 = 0.585). Of the statistically significant hits, proteins with more than a 2-fold increase are circled in red (168 proteins), while those with more than a 2-fold decrease are in blue. A total of 9881 proteins were quantified. n = 4 biological replicates. E, gene ontology analyses of the significantly upregulated proteins with more than 2-fold changes, 168 proteins as described in panel D, are shown. N represents the number of proteins counted in the category. F, 19 quantified proteins in the mevalonate/sterol pathway were curated, and their log2 fold changes are presented as a heat map. dTAG treatment led to the steep depletion of HMGCS1. G, representative families of isoprenylation targets (Rho, Rab, Rheb, and Ras proteins) were curated from 1200 statistically significant hits and plotted as a heat map. FDR, false discovery rate; HMGCR, HMG-CoA reductase; HMGCS1, 3-hydroxy-3-methylglutaryl-coenzyme A synthase 1.

Figure 4

Combinatory targeting HMGCS1 and HMGCR exhibit synergistic antitumor effects.A, the left panel presents a schematic illustration of the combinatorial inhibition of HMGCS1 and HMGCR. In the right panel, the colony formation assay results for HMGCS1-FKBP12^F36V K/I HCT116 cells are shown, where treatments involved either Hymeglusin (5 μM), simvastatin (5 μM), or a combination of both inhibitors. Corresponding inhibitors were applied on days 5 and 7 after seeding, and the colony numbers were counted on day 12. The accompanying quantification graph displays the means ± SD from biological triplicates. B, the left panel illustrates the degradation of HMGCS1 and the inhibition of HMGCR as a combinatorial approach to target the mevalonate pathway. The right panel shows the induction of HMGCS1 following simvastatin treatment, which was completely reversed by dTAGv1 treatment due to the induced degradation in HCT116 cells expressing endogenous HMGCS1-FKBP12^F36V. C, HMGCS1 degradation by dTAGv1 (50 nM) increases the inhibitory effects of simvastatin (4 μM) on the colony-forming activity of HCT116 HMGCS1-FKBP12^F36V cells. D, means ± SD of biological quadruplicate data from panel C. E, HMGCS1 degradation by the dTAG system reduces the IC₅₀ value of simvastatin by 10-fold in HCT116 HMGCS1-FKBP12^F36V knock-in cells (top). Cell viability was assessed 24 h after treating the corresponding cells with increasing concentrations of simvastatin and dTAGv1 (50 nM). The WT HCT116 did not show the synergistic effect of dTAGv1 and simvastatin (bottom). Means ± SD of biological quadruplicate. F, the anticolony formation effect induced by the HMGCS1 degradation and HMGCR inhibition was rescued by supplementation of geranylgeraniol (GGOH, 40 μM). HCT116 HMGCS1-FKBP12^F36V K/I cells were treated with dTAGv1 (50 nM) and simvastatin (4 μM) on days 5 and 7, postseeding. GGOH (40 μM) was added to the media of the corresponding cells on days 5 and 7. Means ± SD of biological triplicate data are shown on the right. G, HMGCS1-FKBP12^F36V K/I HCT116 cells were implanted into nude mice. When the tumor size reached 200 to 250 mm³, we administered two treatment regimens (shown in dotted vertical lines), including saline control, 5 mg/kg simvastatin via gavage three times a week, weekly 5 mg/kg dTAGv1 through i.p., and the combination of both dTAGv1 and simvastatin. Degradation of HMGCS1 by the dTAG system potentiates the tumor growth-suppressing effect of simvastatin. Means ± SEM of five mice per condition are plotted. A paired t test between control and statin + dTAG treatment was performed for days 11, 14, 18, and 21. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. H, survival probability analysis of the mice presented in panel H using Kaplan–Meier software reveals extended mortality in the dTAG and statin cotreatment groups compared to the other control groups. Log-rank (Mantel-Cox) test was performed between the control and statin + dTAG treatment groups. ∗∗p < 0.01. HMGCR, HMG-CoA reductase; HMGCS1, 3-hydroxy-3-methylglutaryl-coenzyme A synthase 1.

Figure 5

Hymeglusin exhibits poor serum stability.A, HCT116, HEK293T, and HepG2 cells were treated with Hymeglusin (0.5 μM) for the indicated time points, followed by lysis and reaction with the HG-FL probe. In-gel fluorescence analysis and immunoblotting were then performed (B) The quantification of relative fluorescence intensity from three replicate experiments in panel A is presented. Mean ± SD. C, the absence of fetal bovine serum (FBS) in the cell culture media delays the appearance of active HMGCS1 with free catalytic cysteine after prolonged treatment of cells with Hymeglusin. D, a workflow to test the effect of FBS on the stability of Hymeglusin. A total of 0.5 μM of Hymeglusin was added to DMEM supplemented with or without 10% FBS. The media were then incubated at 37 °C for 2 or 16 h before being added to HEK293T cells. E, HEK293T cells treated as in panel D were subjected to in-gel fluorescence analysis. F, in vitro incubation of HG-FL with recombinant HMGCS1 (1 μg) or BSA (1 μg) for 1 h produces a green fluorescence signal in the relevant molecular weight regions, indicating a strong interaction between HG-FL and BSA. G, scheme of Ellman’s reagent, which turns the solution into a yellow visible color when reacted with thiol by producing TNB. H, a portion of the samples treated as in (G) were reacted with Ellman’s reagent. Specifically, BSA was incubated with water or chloroacetamide for 10 min, followed by the addition of Ellman’s reagent. The water pretreated sample turned into the expected yellow solution, while the chloroacetamide pretreated BSA solution did not change color, indicating the loss of free thiol due to alkylation. I, the remaining portion of the samples in panels G and H was then incubated with HG-FL, which showed an equal level of green fluorescence signal. BSA, bovine serum albumin; DMEM, Dulbecco’s modified Eagle’s medium; HG-FL, Hymeglusin-fluorescein; HMGCS1, 3-hydroxy-3-methylglutaryl-coenzyme A synthase 1; TNB, 2-nitro-5-thiobenzoate.